1 // Test host codegen.
2 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
3 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
4 // RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
5 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
6 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
7 // RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
8
9 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
10 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
11 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
12 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
13 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
14 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
15 // SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
16
17 // Test target codegen - host bc file has to be created first.
18 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc
19 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s --check-prefix TCHECK --check-prefix TCHECK-64
20 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s
21 // RUN: %clang_cc1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix TCHECK --check-prefix TCHECK-64
22 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc
23 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s --check-prefix TCHECK --check-prefix TCHECK-32
24 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s
25 // RUN: %clang_cc1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix TCHECK --check-prefix TCHECK-32
26
27 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm-bc %s -o %t-ppc-host.bc
28 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck --check-prefix SIMD-ONLY1 %s
29 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o %t %s
30 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY1 %s
31 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm-bc %s -o %t-x86-host.bc
32 // RUN: %clang_cc1 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck --check-prefix SIMD-ONLY1 %s
33 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o %t %s
34 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY1 %s
35 // SIMD-ONLY1-NOT: {{__kmpc|__tgt}}
36
37 // expected-no-diagnostics
38 #ifndef HEADER
39 #define HEADER
40
41 template <typename tx, typename ty>
42 struct TT {
43 tx X;
44 ty Y;
45 };
46
47 // CHECK-DAG: [[TT:%.+]] = type { i64, i8 }
48 // CHECK-DAG: [[TTII:%.+]] = type { i32, i32 }
49 // CHECK-DAG: [[S1:%.+]] = type { double }
50
51 // TCHECK-DAG: [[TT:%.+]] = type { i64, i8 }
52 // TCHECK-DAG: [[TTII:%.+]] = type { i32, i32 }
53 // TCHECK-DAG: [[S1:%.+]] = type { double }
54
55 // CHECK-DAG: [[FP_E:@__omp_offloading_firstprivate_.+_e_l76]] = internal global [[TTII]] zeroinitializer
56 // CHECK-DAG: [[SIZET:@.+]] = private unnamed_addr constant [2 x i{{32|64}}] [i[[SZ:32|64]] 4, i{{64|32}} {{8|4}}]
57 // CHECK-DAG: [[MAPT:@.+]] = private unnamed_addr constant [2 x i64] [i64 288, i64 49]
58 // CHECK-DAG: [[MAPT2:@.+]] = private unnamed_addr constant [9 x i64] [i64 288, i64 161, i64 800, i64 161, i64 161, i64 800, i64 800, i64 161, i64 161]
59 // CHECK-DAG: [[SIZET3:@.+]] = private unnamed_addr constant [2 x i{{32|64}}] [i{{32|64}} 0, i{{32|64}} 8]
60 // CHECK-DAG: [[MAPT3:@.+]] = private unnamed_addr constant [2 x i64] [i64 32, i64 37]
61 // CHECK-DAG: [[MAPT4:@.+]] = private unnamed_addr constant [5 x i64] [i64 547, i64 288, i64 800, i64 800, i64 161]
62 // CHECK-DAG: [[SIZET5:@.+]] = private unnamed_addr constant [3 x i{{32|64}}] [i[[SZ]] 4, i[[SZ]] 1, i[[SZ]] 40]
63 // CHECK-DAG: [[MAPT5:@.+]] = private unnamed_addr constant [3 x i64] [i64 288, i64 288, i64 161]
64 // CHECK-DAG: [[SIZET6:@.+]] = private unnamed_addr constant [2 x i{{32|64}}] [i[[SZ]] 4, i[[SZ]] 40]
65 // CHECK-DAG: [[MAPT6:@.+]] = private unnamed_addr constant [2 x i64] [i64 288, i64 161]
66
67 // CHECK: define {{.*}}[[FOO:@.+]](
foo(int n,double * ptr)68 int foo(int n, double *ptr) {
69 int a = 0;
70 short aa = 0;
71 float b[10];
72 float bn[n];
73 double c[5][10];
74 double cn[5][n];
75 TT<long long, char> d;
76 const TT<int, int> e = {n, n};
77 int *p __attribute__ ((aligned (64))) = &a;
78
79 #pragma omp target firstprivate(a, p)
80 {
81 }
82
83 // a is passed by value to tgt_target
84 // CHECK: [[N_ADDR:%.+]] = alloca i{{[0-9]+}},
85 // CHECK: [[PTR_ADDR:%.+]] = alloca double*,
86 // CHECK: [[A:%.+]] = alloca i{{[0-9]+}},
87 // CHECK: [[A2:%.+]] = alloca i{{[0-9]+}},
88 // CHECK: [[B:%.+]] = alloca [10 x float],
89 // CHECK: [[SSTACK:%.+]] = alloca i8*,
90 // CHECK: [[C:%.+]] = alloca [5 x [10 x double]],
91 // CHECK: [[D:%.+]] = alloca [[TT]],
92 // CHECK: [[P:%.+]] = alloca i32*, align 64
93 // CHECK: [[ACAST:%.+]] = alloca i{{[0-9]+}},
94 // CHECK: [[BASE_PTR_ARR:%.+]] = alloca [2 x i8*],
95 // CHECK: [[PTR_ARR:%.+]] = alloca [2 x i8*],
96 // CHECK: [[A2CAST:%.+]] = alloca i{{[0-9]+}},
97 // CHECK: [[BASE_PTR_ARR2:%.+]] = alloca [9 x i8*],
98 // CHECK: [[PTR_ARR2:%.+]] = alloca [9 x i8*],
99 // CHECK: [[SIZET2:%.+]] = alloca [9 x i{{[0-9]+}}],
100 // CHECK: [[BASE_PTR_ARR3:%.+]] = alloca [2 x i8*],
101 // CHECK: [[PTR_ARR3:%.+]] = alloca [2 x i8*],
102 // CHECK: [[N_ADDR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[N_ADDR]],
103 // CHECK-64: [[N_EXT:%.+]] = zext i{{[0-9]+}} [[N_ADDR_VAL]] to i{{[0-9]+}}
104 // CHECK: [[SSAVE_RET:%.+]] = call i8* @llvm.stacksave()
105 // CHECK: store i8* [[SSAVE_RET]], i8** [[SSTACK]],
106 // CHECK-64: [[BN_VLA:%.+]] = alloca float, i{{[0-9]+}} [[N_EXT]],
107 // CHECK-32: [[BN_VLA:%.+]] = alloca float, i{{[0-9]+}} [[N_ADDR_VAL]],
108 // CHECK: [[N_ADDR_VAL2:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[N_ADDR]],
109 // CHECK-64: [[N_EXT2:%.+]] = zext i{{[0-9]+}} [[N_ADDR_VAL2]] to i{{[0-9]+}}
110 // CHECK-64: [[CN_SIZE:%.+]] = mul{{.+}} i{{[0-9]+}} 5, [[N_EXT2]]
111 // CHECK-32: [[CN_SIZE:%.+]] = mul{{.+}} i{{[0-9]+}} 5, [[N_ADDR_VAL2]]
112 // CHECK: [[CN_VLA:%.+]] = alloca double, i{{[0-9]+}} [[CN_SIZE]],
113 // CHECK: [[AVAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A]],
114 // CHECK-64: [[CONV:%.+]] = bitcast i{{[0-9]+}}* [[ACAST]] to i{{[0-9]+}}*
115 // CHECK-64: store i{{[0-9]+}} [[AVAL]], i{{[0-9]+}}* [[CONV]],
116 // CHECK-32: store i{{[0-9]+}} [[AVAL]], i{{[0-9]+}}* [[ACAST]],
117 // CHECK: [[ACAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[ACAST]],
118 // CHECK: [[P_PTR:%.+]] = load i32*, i32** [[P]], align 64
119 // CHECK: [[BASE_PTR_GEP:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BASE_PTR_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
120 // CHECK: [[ACAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP]] to i{{[0-9]+}}*
121 // CHECK: store i{{[0-9]+}} [[ACAST_VAL]], i{{[0-9]+}}* [[ACAST_TOPTR]],
122 // CHECK: [[PTR_GEP:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTR_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
123 // CHECK: [[ACAST_TOPTR2:%.+]] = bitcast i8** [[PTR_GEP]] to i{{[0-9]+}}*
124 // CHECK: store i{{[0-9]+}} [[ACAST_VAL]], i{{[0-9]+}}* [[ACAST_TOPTR2]],
125 // CHECK: [[BASE_PTR_GEP:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BASE_PTR_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
126 // CHECK: [[PCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP]] to i32***
127 // CHECK: store i32** [[P]], i32*** [[PCAST_TOPTR]],
128 // CHECK: [[PTR_GEP:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTR_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
129 // CHECK: [[PCAST_TOPTR2:%.+]] = bitcast i8** [[PTR_GEP]] to i32**
130 // CHECK: store i32* [[P_PTR]], i32** [[PCAST_TOPTR2]],
131 // CHECK: [[BASE_PTR_GEP_ARG:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BASE_PTR_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
132 // CHECK: [[PTR_GEP_ARG:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTR_ARR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
133 // CHECK: {{.+}} = call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, {{.+}}, i32 2, i8** [[BASE_PTR_GEP_ARG]], i8** [[PTR_GEP_ARG]], i[[SZ]]* getelementptr inbounds ([2 x i[[SZ]]], [2 x i[[SZ]]]* [[SIZET]], i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[MAPT]], i32 0, i32 0), i8** null, i8** null)
134
135 // TCHECK: define weak void @__omp_offloading_{{.+}}(i{{[0-9]+}} [[A_IN:%.+]], i32** nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) [[P_IN:%.+]])
136 // TCHECK: [[A_ADDR:%.+]] = alloca i{{[0-9]+}},
137 // TCHECK: [[P_ADDR:%.+]] = alloca i32**,
138 // TCHECK: [[P_PRIV:%.+]] = alloca i32*,
139 // TCHECK-NOT: alloca i{{[0-9]+}}
140 // TCHECK: store i{{[0-9]+}} [[A_IN]], i{{[0-9]+}}* [[A_ADDR]],
141 // TCHECK: store i32** [[P_IN]], i32*** [[P_ADDR]],
142 // TCHECK-NOT: store i{{[0-9]+}} %
143 // TCHECK: ret void
144
145 #pragma omp target firstprivate(aa, b, bn, c, cn, d)
146 {
147 aa += 1;
148 b[2] = 1.0;
149 bn[3] = 1.0;
150 c[1][2] = 1.0;
151 cn[1][3] = 1.0;
152 d.X = 1;
153 d.Y = 1;
154 }
155
156 // CHECK: [[A2VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A2]],
157 // CHECK: [[A2CASTCONV:%.+]] = bitcast i{{[0-9]+}}* [[A2CAST]] to i{{[0-9]+}}*
158 // CHECK: store i{{[0-9]+}} [[A2VAL]], i{{[0-9]+}}* [[A2CASTCONV]],
159 // CHECK: [[A2CAST_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[A2CAST]],
160 // CHECK-64: [[BN_SIZE:%.+]] = mul{{.+}} i{{[0-9]+}} [[N_EXT]], 4
161 // CHECK-32: [[BN_SZ_SIZE:%.+]] = mul{{.+}} i{{[0-9]+}} [[N_ADDR_VAL]], 4
162 // CHECK-32: [[BN_SIZE:%.+]] = sext i32 [[BN_SZ_SIZE]] to i64
163 // CHECK-64: [[CN_SIZE_1:%.+]] = mul{{.+}} i{{[0-9]+}} 5, [[N_EXT2]]
164 // CHECK-32: [[CN_SIZE_1:%.+]] = mul{{.+}} i{{[0-9]+}} 5, [[N_ADDR_VAL2]]
165 // CHECK-64: [[CN_SIZE_2:%.+]] = mul{{.+}} i{{[0-9]+}} [[CN_SIZE_1]], 8
166 // CHECK-32: [[CN_SZ_SIZE_2:%.+]] = mul{{.+}} i{{[0-9]+}} [[CN_SIZE_1]], 8
167 // CHECK-32: [[CN_SIZE_2:%.+]] = sext i32 [[CN_SZ_SIZE_2]] to i64
168
169 // firstprivate(aa) --> base_ptr = aa, ptr = aa, size = 2 (short)
170 // CHECK: [[BASE_PTR_GEP2_0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
171 // CHECK: [[ACAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_0]] to i{{[0-9]+}}*
172 // CHECK: store i{{[0-9]+}} [[A2CAST_VAL]], i{{[0-9]+}}* [[ACAST_TOPTR]],
173 // CHECK: [[PTR_GEP2_0:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
174 // CHECK: [[ACAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_0]] to i{{[0-9]+}}*
175 // CHECK: store i{{[0-9]+}} [[A2CAST_VAL]], i{{[0-9]+}}* [[ACAST_TOPTR]],
176 // CHECK: [[SIZE_GEPA2:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
177 // CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[SIZE_GEPA2]],
178
179 // firstprivate(b): base_ptr = &b[0], ptr = &b[0], size = 40 (sizeof(float)*10)
180 // CHECK: [[BASE_PTR_GEP2_1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
181 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_1]] to [10 x float]**
182 // CHECK: store [10 x float]* [[B]], [10 x float]** [[BCAST_TOPTR]],
183 // CHECK: [[PTR_GEP2_1:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
184 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_1]] to [10 x float]**
185 // CHECK: store [10 x float]* [[B]], [10 x float]** [[BCAST_TOPTR]],
186 // CHECK: [[SIZE_GEPB:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
187 // CHECK: store i{{[0-9]+}} 40, i{{[0-9]+}}* [[SIZE_GEPB]],
188
189 // firstprivate(bn), 2 entries, n and bn: (1) base_ptr = n, ptr = n, size = 8 ; (2) base_ptr = &c[0], ptr = &c[0], size = n*sizeof(float)
190 // CHECK: [[BASE_PTR_GEP2_2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
191 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_2]] to i{{[0-9]+}}*
192 // CHECK-64: store i{{[0-9]+}} [[N_EXT]], i{{[0-9]+}}* [[BCAST_TOPTR]],
193 // CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL]], i{{[0-9]+}}* [[BCAST_TOPTR]],
194 // CHECK: [[PTR_GEP2_2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
195 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_2]] to i{{[0-9]+}}*
196 // CHECK-64: store i{{[0-9]+}} [[N_EXT]], i{{[0-9]+}}* [[BCAST_TOPTR]],
197 // CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL]], i{{[0-9]+}}* [[BCAST_TOPTR]],
198 // CHECK: [[SIZE_GEPBN_1:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
199 // CHECK: store i{{[0-9]+}} {{[0-9]}}, i{{[0-9]+}}* [[SIZE_GEPBN_1]],
200 // CHECK: [[BASE_PTR_GEP2_3:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
201 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_3]] to float**
202 // CHECK: store float* [[BN_VLA]], float** [[BCAST_TOPTR]],
203 // CHECK: [[SIZE_GEPBN_3:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
204 // CHECK: store i{{[0-9]+}} [[BN_SIZE]], i{{[0-9]+}}* [[SIZE_GEPBN_3]]
205
206 // firstprivate(c): base_ptr = &c[0], ptr = &c[0], size = 400 (5*10*sizeof(double))
207 // CHECK: [[BASE_PTR_GEP2_4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
208 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_4]] to [5 x [10 x double]]**
209 // CHECK: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[BCAST_TOPTR]],
210 // CHECK: [[PTR_GEP2_4:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
211 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_4]] to [5 x [10 x double]]**
212 // CHECK: store [5 x [10 x double]]* [[C]], [5 x [10 x double]]** [[BCAST_TOPTR]],
213 // CHECK: [[SIZE_GEPC_4:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
214 // CHECK: store i{{[0-9]+}} 400, i{{[0-9]+}}* [[SIZE_GEPC_4]],
215
216 // firstprivate(cn), 3 entries, 5, n, cn: (1) base_ptr = 5, ptr = 5, size = 8; (2) (1) base_ptr = n, ptr = n, size = 8; (3) base_ptr = &cn[0], ptr = &cn[0], size = 5*n*sizeof(double)
217 // CHECK: [[BASE_PTR_GEP2_5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 5
218 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_5]] to i{{[0-9]+}}*
219 // CHECK: store i{{[0-9]+}} 5, i{{[0-9]+}}* [[BCAST_TOPTR]],
220 // CHECK: [[PTR_GEP2_5:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 5
221 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_5]] to i{{[0-9]+}}*
222 // CHECK: store i{{[0-9]+}} 5, i{{[0-9]+}}* [[BCAST_TOPTR]],
223 // CHECK: [[SIZE_GEPCN_5:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 5
224 // CHECK: store i{{[0-9]+}} {{[0-9]}}, i{{[0-9]+}}* [[SIZE_GEPCN_5]],
225 // CHECK: [[BASE_PTR_GEP2_6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 6
226 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_6]] to i{{[0-9]+}}*
227 // CHECK-64: store i{{[0-9]+}} [[N_EXT2]], i{{[0-9]+}}* [[BCAST_TOPTR]],
228 // CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL2]], i{{[0-9]+}}* [[BCAST_TOPTR]],
229 // CHECK: [[PTR_GEP2_6:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 6
230 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_6]] to i{{[0-9]+}}*
231 // CHECK-64: store i{{[0-9]+}} [[N_EXT2]], i{{[0-9]+}}* [[BCAST_TOPTR]],
232 // CHECK-32: store i{{[0-9]+}} [[N_ADDR_VAL2]], i{{[0-9]+}}* [[BCAST_TOPTR]],
233 // CHECK: [[SIZE_GEPCN_6:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 6
234 // CHECK: store i{{[0-9]+}} {{[0-9]}}, i{{[0-9]+}}* [[SIZE_GEPCN_6]],
235 // CHECK: [[BASE_PTR_GEP2_7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 7
236 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_7]] to double**
237 // CHECK: store double* [[CN_VLA]], double** [[BCAST_TOPTR]],
238 // CHECK: [[PTR_GEP2_7:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 7
239 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_7]] to double**
240 // CHECK: store double* [[CN_VLA]], double** [[BCAST_TOPTR]],
241 // CHECK: [[SIZE_GEPCN_7:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 7
242 // CHECK: store i{{[0-9]+}} [[CN_SIZE_2]], i{{[0-9]+}}* [[SIZE_GEPCN_7]],
243
244 // firstprivate(d): base_ptr = &d, ptr = &d, size = 16
245 // CHECK: [[BASE_PTR_GEP2_8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 8
246 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP2_8]] to [[TT]]**
247 // CHECK: store [[TT]]* [[D]], [[TT]]** [[BCAST_TOPTR]],
248 // CHECK: [[PTR_GEP2_8:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 8
249 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP2_8]] to [[TT]]**
250 // CHECK: store [[TT]]* [[D]], [[TT]]** [[BCAST_TOPTR]],
251 // CHECK: [[SIZE_GEPCN_8:%.+]] = getelementptr inbounds [9 x i{{[0-9]+}}], [9 x i{{[0-9]+}}]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 8
252 // CHECK: store i{{[0-9]+}} {{[0-9]+}}, i{{[0-9]+}}* [[SIZE_GEPCN_8]],
253
254 // CHECK: [[BASE_PTR_GEP_ARG2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[BASE_PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
255 // CHECK: [[PTR_GEP_ARG2:%.+]] = getelementptr inbounds [9 x i8*], [9 x i8*]* [[PTR_ARR2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
256 // CHECK: [[SIZES_ARG2:%.+]] = getelementptr inbounds [9 x i[[SZ]]], [9 x i[[SZ]]]* [[SIZET2]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
257 // CHECK: {{.+}} = call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, {{.+}}, i32 9, i8** [[BASE_PTR_GEP_ARG2]], i8** [[PTR_GEP_ARG2]], i[[SZ]]* [[SIZES_ARG2]], i64* getelementptr inbounds ([9 x i64], [9 x i64]* [[MAPT2]], i32 0, i32 0), i8** null, i8** null)
258
259 // make sure that firstprivate variables are generated in all cases and that we use those instances for operations inside the
260 // target region
261 // TCHECK: define {{.*}}void @__omp_offloading_{{.+}}(i{{[0-9]+}} [[A2_IN:%.+]], [10 x float]* {{.+}} [[B_IN:%.+]], i{{[0-9]+}} [[BN_SZ:%.+]], float* {{.+}} [[BN_IN:%.+]], [5 x [10 x double]]* {{.+}} [[C_IN:%.+]], i{{[0-9]+}} [[CN_SZ1:%.+]], i{{[0-9]+}} [[CN_SZ2:%.+]], double* {{.+}} [[CN_IN:%.+]], [[TT]]* {{.+}} [[D_IN:%.+]])
262 // TCHECK: [[A2_ADDR:%.+]] = alloca i{{[0-9]+}},
263 // TCHECK: [[B_ADDR:%.+]] = alloca [10 x float]*,
264 // TCHECK: [[VLA_ADDR:%.+]] = alloca i{{[0-9]+}},
265 // TCHECK: [[BN_ADDR:%.+]] = alloca float*,
266 // TCHECK: [[C_ADDR:%.+]] = alloca [5 x [10 x double]]*,
267 // TCHECK: [[VLA_ADDR2:%.+]] = alloca i{{[0-9]+}},
268 // TCHECK: [[VLA_ADDR4:%.+]] = alloca i{{[0-9]+}},
269 // TCHECK: [[CN_ADDR:%.+]] = alloca double*,
270 // TCHECK: [[D_ADDR:%.+]] = alloca [[TT]]*,
271 // TCHECK-NOT: alloca i{{[0-9]+}},
272 // TCHECK: [[B_PRIV:%.+]] = alloca [10 x float],
273 // TCHECK: [[SSTACK:%.+]] = alloca i8*,
274 // TCHECK: [[C_PRIV:%.+]] = alloca [5 x [10 x double]],
275 // TCHECK: [[D_PRIV:%.+]] = alloca [[TT]],
276 // TCHECK: store i{{[0-9]+}} [[A2_IN]], i{{[0-9]+}}* [[A2_ADDR]],
277 // TCHECK: store [10 x float]* [[B_IN]], [10 x float]** [[B_ADDR]],
278 // TCHECK: store i{{[0-9]+}} [[BN_SZ]], i{{[0-9]+}}* [[VLA_ADDR]],
279 // TCHECK: store float* [[BN_IN]], float** [[BN_ADDR]],
280 // TCHECK: store [5 x [10 x double]]* [[C_IN]], [5 x [10 x double]]** [[C_ADDR]],
281 // TCHECK: store i{{[0-9]+}} [[CN_SZ1]], i{{[0-9]+}}* [[VLA_ADDR2]],
282 // TCHECK: store i{{[0-9]+}} [[CN_SZ2]], i{{[0-9]+}}* [[VLA_ADDR4]],
283 // TCHECK: store double* [[CN_IN]], double** [[CN_ADDR]],
284 // TCHECK: store [[TT]]* [[D_IN]], [[TT]]** [[D_ADDR]],
285 // TCHECK: [[CONV_A2ADDR:%.+]] = bitcast i{{[0-9]+}}* [[A2_ADDR]] to i{{[0-9]+}}*
286 // TCHECK: [[B_ADDR_REF:%.+]] = load [10 x float]*, [10 x float]** [[B_ADDR]],
287 // TCHECK: [[BN_SZ_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[VLA_ADDR]],
288 // TCHECK: [[BN_ADDR_REF:%.+]] = load float*, float** [[BN_ADDR]],
289 // TCHECK: [[C_ADDR_REF:%.+]] = load [5 x [10 x double]]*, [5 x [10 x double]]** [[C_ADDR]],
290 // TCHECK: [[CN_SZ1_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[VLA_ADDR2]],
291 // TCHECK: [[CN_SZ2_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[VLA_ADDR4]],
292 // TCHECK: [[CN_ADDR_REF:%.+]] = load double*, double** [[CN_ADDR]],
293 // TCHECK: [[D_ADDR_REF:%.+]] = load [[TT]]*, [[TT]]** [[D_ADDR]],
294
295 // firstprivate(aa): a_priv = a_in
296 // TCHECK-NOT: store i{{[0-9]+}} %
297
298 // firstprivate(b): memcpy(b_priv,b_in)
299 // TCHECK: [[B_PRIV_BCAST:%.+]] = bitcast [10 x float]* [[B_PRIV]] to i8*
300 // TCHECK: [[B_ADDR_REF_BCAST:%.+]] = bitcast [10 x float]* [[B_ADDR_REF]] to i8*
301 // TCHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[B_PRIV_BCAST]], i8* align {{[0-9]+}} [[B_ADDR_REF_BCAST]], {{.+}})
302
303 // TCHECK: [[RET_STACK:%.+]] = call i8* @llvm.stacksave()
304 // TCHECK: store i8* [[RET_STACK]], i8** [[SSTACK]],
305
306 // firstprivate(bn)
307 // TCHECK: [[BN_PRIV:%.+]] = alloca float, i{{[0-9]+}} [[BN_SZ_VAL]],
308 // TCHECK: [[BN_COPY_SZ:%.+]] = mul{{.+}} i{{[0-9]+}} [[BN_SZ_VAL]], 4
309 // TCHECK: [[BN_PRIV__BCAST:%.+]] = bitcast float* [[BN_PRIV]] to i8*
310 // TCHECK: [[BN_REF_IN_BCAST:%.+]] = bitcast float* [[BN_ADDR_REF]] to i8*
311 // TCHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[BN_PRIV__BCAST]], i8* align {{[0-9]+}} [[BN_REF_IN_BCAST]], i{{[0-9]+}} [[BN_COPY_SZ]],{{.+}})
312
313 // firstprivate(c)
314 // TCHECK: [[C_PRIV_BCAST:%.+]] = bitcast [5 x [10 x double]]* [[C_PRIV]] to i8*
315 // TCHECK: [[C_IN_BCAST:%.+]] = bitcast [5 x [10 x double]]* [[C_ADDR_REF]] to i8*
316 // TCHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[C_PRIV_BCAST]], i8* align {{[0-9]+}} [[C_IN_BCAST]],{{.+}})
317
318 // firstprivate(cn)
319 // TCHECK: [[CN_SZ:%.+]] = mul{{.+}} i{{[0-9]+}} [[CN_SZ1_VAL]], [[CN_SZ2_VAL]]
320 // TCHECK: [[CN_PRIV:%.+]] = alloca double, i{{[0-9]+}} [[CN_SZ]],
321 // TCHECK: [[CN_SZ2:%.+]] = mul{{.+}} i{{[0-9]+}} [[CN_SZ1_VAL]], [[CN_SZ2_VAL]]
322 // TCHECK: [[CN_SZ2_CPY:%.+]] = mul{{.+}} i{{[0-9]+}} [[CN_SZ2]], 8
323 // TCHECK: [[CN_PRIV_BCAST:%.+]] = bitcast double* [[CN_PRIV]] to i8*
324 // TCHECK: [[CN_IN_BCAST:%.+]] = bitcast double* [[CN_ADDR_REF]] to i8*
325 // TCHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[CN_PRIV_BCAST]], i8* align {{[0-9]+}} [[CN_IN_BCAST]], i{{[0-9]+}} [[CN_SZ2_CPY]],{{.+}})
326
327 // firstprivate(d)
328 // TCHECK: [[D_PRIV_BCAST:%.+]] = bitcast [[TT]]* [[D_PRIV]] to i8*
329 // TCHECK: [[D_IN_BCAST:%.+]] = bitcast [[TT]]* [[D_ADDR_REF]] to i8*
330 // TCHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[D_PRIV_BCAST]], i8* align {{[0-9]+}} [[D_IN_BCAST]],{{.+}})
331
332 #pragma omp target firstprivate(ptr, e)
333 {
334 ptr[0] = e.X;
335 ptr[0]++;
336 }
337 // CHECK: [[PTR_ADDR_REF:%.+]] = load double*, double** [[PTR_ADDR]],
338
339 // CHECK: [[E_BC:%.+]] = bitcast [[TTII]]* [[E:%.+]] to i8*
340 // CHECK: call void @llvm.memcpy.p0i8.p0i8.i{{64|32}}(i8* {{.*}} bitcast ([[TTII]]* [[FP_E]] to i8*), i8* {{.*}} [[E_BC]], i{{64|32}} 8, i1 false)
341 // CHECK: [[BASE_PTR_GEP3_0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BASE_PTR_ARR3]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
342 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP3_0]] to double**
343 // CHECK: store double* [[PTR_ADDR_REF]], double** [[BCAST_TOPTR]],
344 // CHECK: [[PTR_GEP3_0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTR_ARR3]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
345 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP3_0]] to double**
346 // CHECK: store double* [[PTR_ADDR_REF]], double** [[BCAST_TOPTR]],
347 // CHECK: [[BASE_PTR_GEP3_1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BASE_PTR_ARR3]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
348 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTR_GEP3_1]] to [[TTII]]**
349 // CHECK: store [[TTII]]* [[FP_E]], [[TTII]]** [[BCAST_TOPTR]],
350 // CHECK: [[PTR_GEP3_1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTR_ARR3]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
351 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTR_GEP3_1]] to [[TTII]]**
352 // CHECK: store [[TTII]]* [[FP_E]], [[TTII]]** [[BCAST_TOPTR]],
353
354 // CHECK: [[BASE_PTR_GEP_ARG3:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BASE_PTR_ARR3]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
355 // CHECK: [[PTR_GEP_ARG3:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTR_ARR3]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
356 // CHECK: {{.+}} = call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, {{.+}}, i32 2, i8** [[BASE_PTR_GEP_ARG3]], i8** [[PTR_GEP_ARG3]], i[[SZ]]* getelementptr inbounds ([2 x i[[SZ]]], [2 x i[[SZ]]]* [[SIZET3]], i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[MAPT3]], i32 0, i32 0), i8** null, i8** null)
357
358 // TCHECK: define weak void @__omp_offloading_{{.+}}(double* [[PTR_IN:%.+]], [[TTII]]* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) [[E:%.+]])
359 // TCHECK-NOT: alloca [[TTII]],
360 // TCHECK: [[PTR_ADDR:%.+]] = alloca double*,
361 // TCHECK-NOT: alloca [[TTII]],
362 // TCHECK-NOT: alloca double*,
363 // TCHECK: store double* [[PTR_IN]], double** [[PTR_ADDR]],
364 // TCHECK-NOT: store double* %
365
366 return a;
367 }
368
369 template <typename tx>
ftemplate(int n)370 tx ftemplate(int n) {
371 tx a = 0;
372 tx b[10];
373
374 #pragma omp target firstprivate(a, b)
375 {
376 a += 1;
377 b[2] += 1;
378 }
379
380 return a;
381 }
382
fstatic(int n)383 static int fstatic(int n) {
384 int a = 0;
385 char aaa = 0;
386 int b[10];
387
388 #pragma omp target firstprivate(a, aaa, b)
389 {
390 a += 1;
391 aaa += 1;
392 b[2] += 1;
393 }
394
395 return a;
396 }
397
398 // TCHECK: define weak void @__omp_offloading_{{.+}}(i{{[0-9]+}} [[A_IN:%.+]], i{{[0-9]+}} [[A3_IN:%.+]], [10 x i{{[0-9]+}}]*{{.+}} [[B_IN:%.+]])
399 // TCHECK: [[A_ADDR:%.+]] = alloca i{{[0-9]+}},
400 // TCHECK: [[A3_ADDR:%.+]] = alloca i{{[0-9]+}},
401 // TCHECK: [[B_ADDR:%.+]] = alloca [10 x i{{[0-9]+}}]*,
402 // TCHECK-NOT: alloca i{{[0-9]+}},
403 // TCHECK: [[B_PRIV:%.+]] = alloca [10 x i{{[0-9]+}}],
404 // TCHECK: store i{{[0-9]+}} [[A_IN]], i{{[0-9]+}}* [[A_ADDR]],
405 // TCHECK: store i{{[0-9]+}} [[A3_IN]], i{{[0-9]+}}* [[A3_ADDR]],
406 // TCHECK: store [10 x i{{[0-9]+}}]* [[B_IN]], [10 x i{{[0-9]+}}]** [[B_ADDR]],
407 // TCHECK-64: [[A_CONV:%.+]] = bitcast i{{[0-9]+}}* [[A_ADDR]] to i{{[0-9]+}}*
408 // TCHECK: [[A3_CONV:%.+]] = bitcast i{{[0-9]+}}* [[A3_ADDR]] to i8*
409 // TCHECK: [[B_ADDR_REF:%.+]] = load [10 x i{{[0-9]+}}]*, [10 x i{{[0-9]+}}]** [[B_ADDR]],
410
411 // firstprivate(a): a_priv = a_in
412
413 // firstprivate(aaa)
414 // TCHECK-NOT: store i{{[0-9]+}} %
415
416 // firstprivate(b)
417 // TCHECK: [[B_PRIV_BCAST:%.+]] = bitcast [10 x i{{[0-9]+}}]* [[B_PRIV]] to i8*
418 // TCHECK: [[B_IN_BCAST:%.+]] = bitcast [10 x i{{[0-9]+}}]* [[B_ADDR_REF]] to i8*
419 // TCHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[B_PRIV_BCAST]], i8* align {{[0-9]+}} [[B_IN_BCAST]],{{.+}})
420
421 // TCHECK: ret void
422
423 struct S1 {
424 double a;
425
r1S1426 int r1(int n) {
427 int b = n + 1;
428 short int c[2][n];
429
430 #pragma omp target firstprivate(b, c)
431 {
432 this->a = (double)b + 1.5;
433 c[1][1] = ++a;
434 }
435
436 return c[1][1] + (int)b;
437 }
438
439 // on the host side, we first generate r1, then the static function and the template above
440 // CHECK: define{{.+}} i32 {{.+}}([[S1]]* {{.+}}, i{{[0-9]+}} {{.+}})
441 // CHECK: [[BASE_PTRS4:%.+]] = alloca [5 x i8*],
442 // CHECK: [[PTRS4:%.+]] = alloca [5 x i8*],
443 // CHECK: [[SIZET4:%.+]] = alloca [5 x i{{[0-9]+}}],
444
445 // map(this: this ptr is implicitly captured (not firstprivate matter)
446 // CHECK: [[BP0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
447 // CHECK: [[CBP0:%.+]] = bitcast i8** [[BP0]] to %struct.S1**
448 // CHECK: store %struct.S1* [[THIS:%.+]], %struct.S1** [[CBP0]],
449 // CHECK: [[P0:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
450 // CHECK: [[CP0:%.+]] = bitcast i8** [[P0]] to double**
451 // CHECK: store double* [[A:%.+]], double** [[CP0]],
452 // CHECK: [[SZ0:%.+]] = getelementptr inbounds [5 x i{{[0-9]+}}], [5 x i{{[0-9]+}}]* [[SIZET4]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
453 // CHECK: store i{{[0-9]+}} 8, i{{[0-9]+}}* [[SZ0]],
454
455 // firstprivate(b): base_ptr = b, ptr = b, size = 4 (pass by-value)
456 // CHECK: [[BASE_PTRS_GEP4_1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
457 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_1]] to i{{[0-9]+}}*
458 // CHECK: store i{{[0-9]+}} [[B_CAST:%.+]], i{{[0-9]+}}* [[BCAST_TOPTR]],
459 // CHECK: [[PTRS_GEP4_1:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
460 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_1]] to i{{[0-9]+}}*
461 // CHECK: store i{{[0-9]+}} [[B_CAST]], i{{[0-9]+}}* [[BCAST_TOPTR]],
462 // CHECK: [[SIZES_GEP4_1:%.+]] = getelementptr inbounds [5 x i{{[0-9]+}}], [5 x i{{[0-9]+}}]* [[SIZET4]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
463 // CHECK: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIZES_GEP4_1]],
464
465 // firstprivate(c), 3 entries: 2, n, c
466 // CHECK: [[BASE_PTRS_GEP4_2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
467 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_2]] to i{{[0-9]+}}*
468 // CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[BCAST_TOPTR]],
469 // CHECK: [[PTRS_GEP4_2:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
470 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_2]] to i{{[0-9]+}}*
471 // CHECK: store i{{[0-9]+}} 2, i{{[0-9]+}}* [[BCAST_TOPTR]],
472 // CHECK: [[SIZES_GEP4_2:%.+]] = getelementptr inbounds [5 x i{{[0-9]+}}], [5 x i{{[0-9]+}}]* [[SIZET4]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
473 // CHECK-64: store i{{[0-9]+}} 8, i{{[0-9]+}}* [[SIZES_GEP4_2]],
474 // CHECK-32: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIZES_GEP4_2]],
475 // CHECK: [[BASE_PTRS_GEP4_3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
476 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_3]] to i{{[0-9]+}}*
477 // CHECK: store i{{[0-9]+}} [[N:%.+]], i{{[0-9]+}}* [[BCAST_TOPTR]],
478 // CHECK: [[PTRS_GEP4_3:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
479 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_3]] to i{{[0-9]+}}*
480 // CHECK: store i{{[0-9]+}} [[N]], i{{[0-9]+}}* [[BCAST_TOPTR]],
481 // CHECK: [[SIZES_GEP4_3:%.+]] = getelementptr inbounds [5 x i{{[0-9]+}}], [5 x i{{[0-9]+}}]* [[SIZET4]], i{{[0-9]+}} 0, i{{[0-9]+}} 3
482 // CHECK-64: store i{{[0-9]+}} 8, i{{[0-9]+}}* [[SIZES_GEP4_3]],
483 // CHECK-32: store i{{[0-9]+}} 4, i{{[0-9]+}}* [[SIZES_GEP4_3]],
484 // CHECK: [[BASE_PTRS_GEP4_4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[BASE_PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
485 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP4_4]] to i{{[0-9]+}}**
486 // CHECK: store i{{[0-9]+}}* [[B:%.+]], i{{[0-9]+}}** [[BCAST_TOPTR]],
487 // CHECK: [[PTRS_GEP4_4:%.+]] = getelementptr inbounds [5 x i8*], [5 x i8*]* [[PTRS4]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
488 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP4_4]] to i{{[0-9]+}}**
489 // CHECK: store i{{[0-9]+}}* [[B]], i{{[0-9]+}}** [[BCAST_TOPTR]],
490 // CHECK: [[SIZES_GEP4_4:%.+]] = getelementptr inbounds [5 x i{{[0-9]+}}], [5 x i{{[0-9]+}}]* [[SIZET4]], i{{[0-9]+}} 0, i{{[0-9]+}} 4
491 // CHECK: store i{{[0-9]+}} [[B_SIZE:%.+]], i{{[0-9]+}}* [[SIZES_GEP4_4]],
492
493 // only check that we use the map types stored in the global variable
494 // CHECK: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, {{.+}}, i32 5, i8** {{.+}}, i8** {{.+}}, i{{[0-9]+}}* {{.+}}, i64* getelementptr inbounds ([5 x i64], [5 x i64]* [[MAPT4]], i32 0, i32 0), i8** null, i8** null)
495
496 // TCHECK: define weak void @__omp_offloading_{{.+}}([[S1]]* [[TH:%.+]], i{{[0-9]+}} [[B_IN:%.+]], i{{[0-9]+}} [[VLA:%.+]], i{{[0-9]+}} [[VLA1:%.+]], i{{[0-9]+}}{{.+}} [[C_IN:%.+]])
497 // TCHECK: [[TH_ADDR:%.+]] = alloca [[S1]]*,
498 // TCHECK: [[B_ADDR:%.+]] = alloca i{{[0-9]+}},
499 // TCHECK: [[VLA_ADDR:%.+]] = alloca i{{[0-9]+}},
500 // TCHECK: [[VLA_ADDR2:%.+]] = alloca i{{[0-9]+}},
501 // TCHECK: [[C_ADDR:%.+]] = alloca i{{[0-9]+}}*,
502 // TCHECK-NOT: alloca i{{[0-9]+}},
503 // TCHECK: [[SSTACK:%.+]] = alloca i8*,
504
505 // TCHECK: store [[S1]]* [[TH]], [[S1]]** [[TH_ADDR]],
506 // TCHECK: store i{{[0-9]+}} [[B_IN]], i{{[0-9]+}}* [[B_ADDR]],
507 // TCHECK: store i{{[0-9]+}} [[VLA]], i{{[0-9]+}}* [[VLA_ADDR]],
508 // TCHECK: store i{{[0-9]+}} [[VLA1]], i{{[0-9]+}}* [[VLA_ADDR2]],
509 // TCHECK: store i{{[0-9]+}}* [[C_IN]], i{{[0-9]+}}** [[C_ADDR]],
510 // TCHECK: [[TH_ADDR_REF:%.+]] = load [[S1]]*, [[S1]]** [[TH_ADDR]],
511 // TCHECK-64: [[B_ADDR_CONV:%.+]] = bitcast i{{[0-9]+}}* [[B_ADDR]] to i{{[0-9]+}}*
512 // TCHECK: [[VLA_ADDR_REF:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[VLA_ADDR]],
513 // TCHECK: [[VLA_ADDR_REF2:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[VLA_ADDR2]],
514 // TCHECK: [[C_ADDR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[C_ADDR]],
515
516 // firstprivate(b)
517 // TCHECK-NOT: store i{{[0-9]+}} %
518
519 // TCHECK: [[RET_STACK:%.+]] = call i8* @llvm.stacksave()
520 // TCHECK: store i8* [[RET_STACK:%.+]], i8** [[SSTACK]],
521
522 // firstprivate(c)
523 // TCHECK: [[C_SZ:%.+]] = mul{{.+}} i{{[0-9]+}} [[VLA_ADDR_REF]], [[VLA_ADDR_REF2]]
524 // TCHECK: [[C_PRIV:%.+]] = alloca i{{[0-9]+}}, i{{[0-9]+}} [[C_SZ]],
525 // TCHECK: [[C_SZ2:%.+]] = mul{{.+}} i{{[0-9]+}} [[VLA_ADDR_REF]], [[VLA_ADDR_REF2]]
526 // TCHECK: [[C_SZ_CPY:%.+]] = mul{{.+}} i{{[0-9]+}} [[C_SZ2]], 2
527 // TCHECK: [[C_PRIV_BCAST:%.+]] = bitcast i{{[0-9]+}}* [[C_PRIV]] to i8*
528 // TCHECK: [[C_IN_BCAST:%.+]] = bitcast i{{[0-9]+}}* [[C_ADDR_REF]] to i8*
529 // TCHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[C_PRIV_BCAST]], i8* align {{[0-9]+}} [[C_IN_BCAST]],{{.+}})
530
531 // finish
532 // TCHECK: [[RELOAD_SSTACK:%.+]] = load i8*, i8** [[SSTACK]],
533 // TCHECK: call void @llvm.stackrestore(i8* [[RELOAD_SSTACK]])
534 // TCHECK: ret void
535
536 // static host function
537 // CHECK: define{{.+}} i32 {{.+}}(i{{[0-9]+}} {{.+}})
538 // CHECK: [[BASE_PTRS5:%.+]] = alloca [3 x i8*],
539 // CHECK: [[PTRS5:%.+]] = alloca [3 x i8*],
540
541 // firstprivate(a): by value
542 // CHECK: [[BASE_PTRS_GEP5_0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BASE_PTRS5]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
543 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP5_0]] to i{{[0-9]+}}*
544 // CHECK: store i{{[0-9]+}} [[A_CAST:%.+]], i{{[0-9]+}}* [[BCAST_TOPTR]],
545 // CHECK: [[PTRS_GEP5_0:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PTRS5]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
546 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP5_0]] to i{{[0-9]+}}*
547 // CHECK: store i{{[0-9]+}} [[A_CAST]], i{{[0-9]+}}* [[BCAST_TOPTR]],
548
549 // firstprivate(aaa): by value
550 // CHECK: [[BASE_PTRS_GEP5_1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BASE_PTRS5]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
551 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP5_1]] to i{{[0-9]+}}*
552 // CHECK: store i{{[0-9]+}} [[A3_CAST:%.+]], i{{[0-9]+}}* [[BCAST_TOPTR]],
553 // CHECK: [[PTRS_GEP5_1:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PTRS5]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
554 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP5_1]] to i{{[0-9]+}}*
555 // CHECK: store i{{[0-9]+}} [[A3_CAST]], i{{[0-9]+}}* [[BCAST_TOPTR]],
556
557 // firstprivate(b): base_ptr = &b[0], ptr= &b[0]
558 // CHECK: [[BASE_PTRS_GEP5_2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[BASE_PTRS5]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
559 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP5_2]] to [10 x i{{[0-9]+}}]**
560 // CHECK: store [10 x i{{[0-9]+}}]* [[B:%.+]], [10 x i{{[0-9]+}}]** [[BCAST_TOPTR]],
561 // CHECK: [[PTRS_GEP5_2:%.+]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[PTRS5]], i{{[0-9]+}} 0, i{{[0-9]+}} 2
562 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP5_2]] to [10 x i{{[0-9]+}}]**
563 // CHECK: store [10 x i{{[0-9]+}}]* [[B]], [10 x i{{[0-9]+}}]** [[BCAST_TOPTR]],
564
565 // only check that the right sizes and map types are used
566 // CHECK: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, {{.+}}, i32 3, i8** {{.+}}, i8** {{.+}}, i[[SZ]]* getelementptr inbounds ([3 x i[[SZ]]], [3 x i[[SZ]]]* [[SIZET5]], i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* [[MAPT5]], i32 0, i32 0), i8** null, i8** null)
567 };
568
bar(int n,double * ptr)569 int bar(int n, double *ptr) {
570 int a = 0;
571 a += foo(n, ptr);
572 S1 S;
573 a += S.r1(n);
574 a += fstatic(n);
575 a += ftemplate<int>(n);
576
577 return a;
578 }
579
580 // template host and device
581
582 // CHECK: define{{.+}} i32 {{.+}}(i{{[0-9]+}} {{.+}})
583 // CHECK: [[BASE_PTRS6:%.+]] = alloca [2 x i8*],
584 // CHECK: [[PTRS6:%.+]] = alloca [2 x i8*],
585
586 // firstprivate(a): by value
587 // CHECK: [[BASE_PTRS_GEP6_0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BASE_PTRS6]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
588 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP6_0]] to i{{[0-9]+}}*
589 // CHECK: store i{{[0-9]+}} [[AT_CAST:%.+]], i{{[0-9]+}}* [[BCAST_TOPTR]],
590 // CHECK: [[PTRS_GEP6_0:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS6]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
591 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP6_0]] to i{{[0-9]+}}*
592 // CHECK: store i{{[0-9]+}} [[AT_CAST]], i{{[0-9]+}}* [[BCAST_TOPTR]],
593
594 // firstprivate(b): pointer
595 // CHECK: [[BASE_PTRS_GEP6_1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[BASE_PTRS6]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
596 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[BASE_PTRS_GEP6_1]] to [10 x i{{[0-9]+}}]**
597 // CHECK: store [10 x i{{[0-9]+}}]* [[B:%.+]], [10 x i{{[0-9]+}}]** [[BCAST_TOPTR]],
598 // CHECK: [[PTRS_GEP6_1:%.+]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[PTRS6]], i{{[0-9]+}} 0, i{{[0-9]+}} 1
599 // CHECK: [[BCAST_TOPTR:%.+]] = bitcast i8** [[PTRS_GEP6_1]] to [10 x i{{[0-9]+}}]**
600 // CHECK: store [10 x i{{[0-9]+}}]* [[B]], [10 x i{{[0-9]+}}]** [[BCAST_TOPTR]],
601
602 // CHECK: call i32 @__tgt_target_mapper(%struct.ident_t* @{{.+}}, i64 -1, {{.+}}, i32 2, i8** {{.+}}, i8** {{.+}}, i[[SZ]]* getelementptr inbounds ([2 x i[[SZ]]], [2 x i[[SZ]]]* [[SIZET6]], i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* [[MAPT6]], i32 0, i32 0), i8** null, i8** null)
603
604 // TCHECK: define weak void @__omp_offloading_{{.+}}(i{{[0-9]+}} [[A_IN:%.+]], [10 x i{{[0-9]+}}]*{{.+}} [[B_IN:%.+]])
605 // TCHECK: [[A_ADDR:%.+]] = alloca i{{[0-9]+}},
606 // TCHECK: [[B_ADDR:%.+]] = alloca [10 x i{{[0-9]+}}]*,
607 // TCHECK-NOT: alloca i{{[0-9]+}},
608 // TCHECK: [[B_PRIV:%.+]] = alloca [10 x i{{[0-9]+}}],
609 // TCHECK: store i{{[0-9]+}} [[A_IN]], i{{[0-9]+}}* [[A_ADDR]],
610 // TCHECK: store [10 x i{{[0-9]+}}]* [[B_IN]], [10 x i{{[0-9]+}}]** [[B_ADDR]],
611 // TCHECK-64: [[A_ADDR_CONV:%.+]] = bitcast i{{[0-9]+}}* [[A_ADDR]] to i{{[0-9]+}}*
612 // TCHECK: [[B_ADDR_REF:%.+]] = load [10 x i{{[0-9]+}}]*, [10 x i{{[0-9]+}}]** [[B_ADDR]],
613
614 // firstprivate(a)
615 // TCHECK-NOT: store i{{[0-9]+}} %
616
617 // firstprivate(b)
618 // TCHECK: [[B_PRIV_BCAST:%.+]] = bitcast [10 x i{{[0-9]+}}]* [[B_PRIV]] to i8*
619 // TCHECK: [[B_IN_BCAST:%.+]] = bitcast [10 x i{{[0-9]+}}]* [[B_ADDR_REF]] to i8*
620 // TCHECK: call void @llvm.memcpy.{{.+}}(i8* align {{[0-9]+}} [[B_PRIV_BCAST]], i8* align {{[0-9]+}} [[B_IN_BCAST]],{{.+}})
621
622 // TCHECK: ret void
623
624 #endif
625